US5692012A - Method for image compression coding in an image transmission system - Google Patents

Method for image compression coding in an image transmission system Download PDF

Info

Publication number
US5692012A
US5692012A US08/341,534 US34153495A US5692012A US 5692012 A US5692012 A US 5692012A US 34153495 A US34153495 A US 34153495A US 5692012 A US5692012 A US 5692012A
Authority
US
United States
Prior art keywords
block
sub
blocks
image
codebook
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
US08/341,534
Other languages
English (en)
Inventor
Jorma Virtamo
Seppo Valli
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
F Poszat HU LLC
Original Assignee
Valtion Teknillinen Tutkimuskeskus
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Valtion Teknillinen Tutkimuskeskus filed Critical Valtion Teknillinen Tutkimuskeskus
Assigned to VALTION TEKNILLINEN TUTKIMUSKESKUS reassignment VALTION TEKNILLINEN TUTKIMUSKESKUS ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: VALLI, SEPPO, VIRTAMO, JORMA
Application granted granted Critical
Publication of US5692012A publication Critical patent/US5692012A/en
Assigned to PALUS A16, LLC reassignment PALUS A16, LLC ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: VALTION TEKNILLINEN TUTKIMUSKESKUS
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T9/00—Image coding
    • G06T9/008—Vector quantisation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/94—Vector quantisation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/96—Tree coding, e.g. quad-tree coding
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/30—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability

Definitions

  • the invention relates to the method defined in the introductory section of patent claim 1 for image compression coding in an image transmission system, particularly for realizing compression coding in a digital image transmission system.
  • VQ vector quantization
  • the VQ method is discussed extensively in the reference publication /1/.
  • the image to be transmitted is divided into blocks, and for each block, there is searched in the codebook a codevector that best corresponds to the particular block.
  • the vector index of this codevector i.e. its location in the codebook, is transmitted.
  • the said vector is looked up in a corresponding codebook on the basis of the vector index of the codevector.
  • the transmitted image is reconstructed, block by block, by means of the said codevectors.
  • the codebook is designed beforehand by using training images and a suitable codebook design algorithm. In a corresponding fashion, the training images are divided down to the used block size while creating the codebook.
  • a drawback of the said VQ method is its limited block size.
  • the efficiency of the process stems from the fact that all statistical dependencies within the block are taken into account.
  • the correlations over the block boundaries remain unexploited.
  • large scale correlations can be utilized to achieve a higher compression only by using a larger block size.
  • This leads to an exponential growth of the codebook size for a fixed picture quality.
  • the number of comparisons to find the best codevector in the codebook also grows respectively. Therefore VQ becomes computationally unpractical with large block sizes.
  • the block size is limited to the order of 4 ⁇ 4 pixels, which leaves a lot of statistical dependencies beyond the reach of the method.
  • TSVQ tree search VQ
  • image quality terms TSVQ works slightly worse than the standard VQ.
  • VQ Another modified vector quantization process, the cascaded VQ, represents a direct attack on the main problem of standard VQ, i.e. the problem with block size.
  • a relatively large block size is chosen in order to capture the large scale correlations.
  • the employed codebooks are reasonable in size and thus easily manageable.
  • the method is applied in successive coding stages. At each of the later stages, the coding error of the previous stage is quantized.
  • a drawback is that any correlations that may exist between successive codevectors are lost.
  • a third modified vector quantization process which is a refinement of the cascaded VQ
  • the block size is changed when proceeding from the first coding stage to the next.
  • the coding error which contains the fine details of the image, is coded using sub-blocks.
  • the classified VQ the blocks are classified prior to coding.
  • the classification is based on local image features. This improves especially the representation of edges, which is found to be rather poor in standard VQ.
  • the so called finite-state VQ FSVQ
  • inter-block correlations are utilized on the basis of the above mentioned classification.
  • the class of the block is defined with the aid of the larger image context.
  • the image context includes a number of neighbouring blocks that have already been coded and transmitted.
  • classification errors may reduce the quality of the coding output of the edges represented in the image.
  • a sixth modified vector quantization process utilizes the joint distribution of the vector indices of a group of blocks. Special codewords saved in a so called address codebook are reserved for the most probable combinations of vector indices. For all other combinations, the vector indices are transmitted separately. In updating the address codebook, there are utilized the detected joint probabilities of different combinations of vector indices between neighbouring blocks.
  • the address codebook is formed as a distortionless post coding scheme on top of standard VQ, and gives the same picture quality at a lower bit rate than standard VQ.
  • the object of the present invention is to introduce a new method for image compression coding in an image transmission system, whereby the block size problem can be avoided at least to some extent.
  • Another object of the invention is to achieve a new compression coding method, whereby problems connected to modified vector quantization processes can be overcome.
  • the method of the invention for image compression coding in an image transmission system particularly for carrying out image compression in a digital image transmission system
  • a VQ process where the image is divided into blocks at the transmitting end, and for each block there is searched in the codebook a best corresponding codevector, and the vector index of this codevector, i.e. its location in the codebook, is transmitted to the receiving end, where the respective codevector is looked up in an identical codebook on the basis of the index, and the image is reconstructed block by block with the aid of the codevectors;
  • the said codebook is designed beforehand by means of training images, which images are in similar fashion divided into blocks according to the employed block size.
  • the block to be coded is divided into quadrants or sub-blocks
  • each sub-block is subjected to vector quantization by a classified codebook, and the class thereof is defined on the basis of the vector index of the original block and the quadrant label of the sub-block;
  • a classified codebook which is chosen on the basis of the vector index of the transmitted original block and the quadrant label of the sub-block;
  • the codevector in question is looked up in the chosen classified codebook which is identical with the classified codebook at the transmitting end, on the basis of the vector index of the transmitted sub-block, so that a reconstruction is obtained for the sub-block; and in the method of the invention, while creating each classified codebook, the employed training image material is that part of the original training image set which belongs to the respective class, i.e. those quadrant-size sub-blocks of the training image set which receive a class index corresponding to the particular class of the codebook, when the above described method is applied to the training image set itself.
  • vector quantization is applied successively, starting with a suitably chosen block size and by proceeding to its quadrants.
  • Vector quantization carried out at the latter stage utilizes information that is already coded using a larger block size by means of the so called classification.
  • the employed defining index of a class is the vector index obtained from the first quantization, as well as the quadrant label.
  • a quadrant i.e. a sub-block
  • quadrant sub-blocks are further divided into quadrant sub-blocks, and these further into smaller and smaller sub-blocks by following the same principle, so that each quadrant and sub-block is coded with a classified codebook, where the classifiers are the vector indices of all the parent blocks of the said quadrant or sub-block whereto the quadrant or sub-block to be coded belongs, as well as the quandrant labels of the concerned quadrant or sub-block and the said parent blocks, and where the classified codebooks are designed beforehand in the above described fashion.
  • vector quantization is applied by proceeding from a suitably chosen block, i.e. block size, to its quadrants, then further down to their quadrants and so on, as far as the desired depth.
  • a suitably chosen block i.e. block size
  • vector quantization in which the information that was already coded in the larger block size is utilized by means of classification; the defining indices of the class are vector indices obtained as results of earlier quantizations and corresponding quadrant labels. Coding is continued down to the desired depth, as far as the bitrate of the employed transmission channel allows, or until the picture quality is otherwise sufficient.
  • the image is thus sharpened step by step, by the aid of relatively little extra information.
  • the image block is reconstructed on the basis of the information coded so far, and there is formed a difference between the original image block and the reconstructed image block, which difference block thus contains the coding errors of the previously performed coding step, and the method of the invention is applied to the thus created difference block instead of the image block.
  • the difference block is divided into quadrant blocks, which are vector quantized by utilizing the information that was already coded in larger block size by means of classification.
  • the classified codebooks are designed, according to the method of the invention, by using as training material those sub-blocks of the difference block that obtain corresponding class indices as the considered sub-block of the difference block, when the method is applied to the training image set itself.
  • An advantage of the above described embodiment is that the difference signal is more compact, i.e. the values contained in the difference block fluctuate within a more limited range than the values of the image signal proper, which makes coding more efficient.
  • the information coded with larger block size gives a poorer description of the features of the difference block, and this decreases the advantage and coding efficiency that are characteristic of the method of the invention.
  • the realization of an application that codes the difference signal also is somewhat more complicated.
  • the coding is made more effective by separating, prior to the coding, the mean value of the gray level values of the pixels contained in each block of the image, and by transmitting the mean value separately from the transmitting end to the receiving end.
  • the image is subjected to low pass filtering at the transmitting end, and the filtered image is separately transmitted to the receiving end, and one of the above described methods or an application thereof is applied only to the high-frequency component of the image, i.e. to the difference image of the original image and the low pass filtered image.
  • the low-frequency component contains the most important part of the image, and little information and little channel capacity are needed for its transmission. Thus the blockiness of the image is reduced, particularly in "smooth" areas, i.e. in areas containing few changes.
  • the sizes of the codebooks used in coding sub-blocks and possibly subsequent smaller sub-blocks are varied case by case, on the basis of the image content of the coded block, so that in coding blocks with few details, there are used smaller codebooks than in coding blocks with a lot of details, in which case the information of the size of the codebook is transmitted as extra information.
  • the sizes of the codebooks used in coding sub-blocks and possibly subsequent smaller sub-blocks are varied on the basis of the image content of the coded block, so that in coding blocks with few details, there are used smaller codebooks than in coding blocks with a lot of details, and the size of the codebook is defined permanently on the basis of the class index. In this case additional information is not transmitted, but the choice takes place automatically.
  • the method of the invention utilizes the large-scale correlations of the image. This is based on the suitably chosen large initial block size. With respect to accuracy, the process can be modified in a desired fashion, because the image is coded in several stages starting with a relatively large block, which is subsequently divided into smaller and smaller sub-blocks. The more sub-blocks are coded, the higher the accuracy of the image in the respective area becomes.
  • Another advantage of the invention is that it is extremely efficient and the subjective quality of the coded images is good.
  • Yet another advantage of the invention is that the method is fairly easy to apply, and low transmission rates can be achieved.
  • FIG. 1 is a schematical illustration of a vector coding method
  • FIGS. 2a, 2b and 2c are schematical illustrations of how an image block is divided into sub-blocks in the image compression coding method of the invention
  • FIG. 3 is a block diagram illustrating a coder according to the compression coding method of the invention.
  • FIG. 4 respectively illustrates a decoder
  • FIG. 5 illustrates another coder according to the compression coding method of the invention.
  • the principle of vector quantization is illustrated in the block diagram of FIG. 1.
  • An image 1, obtained from a camera such as a video camera, is recorded in the image memory 2.
  • the image 1 stored in the image memory 2 is divided into blocks 4, i.e. vectors.
  • a typical size for the block 4 is 4 ⁇ 4 pixels.
  • Each pixel contains the gray tone information on a predetermined gray scale.
  • the processing unit 3 for each block 4 of the image there is searched the best corresponding codevector 7a from the codebook 6 stored in the codebook memory 5.
  • the codebook 6 is designed beforehand by means of training images and by using a suitable codebook design algorithm, for instance the so called LBG algorithm /2/.
  • the order number a of the codevector 7a in the codebook 6, i.e. its vector index, is transmitted through the transmission channel 8 from the transmitting unit A to the receiving unit B.
  • the image is reconstructed in the image memory 9 of the receiving unit B by looking up, by means of the image processing unit 10, the codevector 7a corresponding to the vector index a in the codebook 12 recorded in the memory 11 of the codebook.
  • the image can be displayed on a monitor.
  • the method of the invention for image compression coding in an image transmission system utilizes the above described vector quantization process. This is illustrated by the block diagram of FIG. 3.
  • a larger block size than in standard VQ is chosen.
  • the size of the block 13 (cf. FIG. 2a) is for instance 8 ⁇ 8 pixels.
  • the block is coded in the first sub-coder 16 by means of a first codebook 17 with a relatively small number of codevectors.
  • the number of the first codevectors 18 is for instance 64.
  • the size of the codevector 18 corresponds to the size of the block 13 and is for example 8 ⁇ 8 pixels. This means that at this coding stage, the image can be figured out only roughly, i.e. only the basic features contained in the low-frequency components of the image can be identified.
  • each sub-block 14 is coded by means of the second codebook 20, which is compiled of a number of classified codebooks, by using as classifiers the vector index i of the original block 13 and the quadrant label A', B', C' and D' of the sub-block 14, i.e. its location inside the original block 13.
  • the sub-block 14 there is utilized, from among the classified codebooks 20, that particular codebook which is defined by the label of each quadrant 14 of the original block 13, and by the vector index i of each original block 13.
  • the first codevector 18 found on the basis of the original block 13 defines the rough features of the image in each block thereof, and the second codevectors 21 from the second codebooks 20 sharpen the image as for the sub-blocks 14.
  • the size of the second codevector 21 corresponds to the size of the sub-block 14 and is for instance 4 ⁇ 4 pixels.
  • the method of the invention can be further applied by assuming at the following stage each of the sub-blocks 14A', 14B', 14C', and 14D', obtained in the above described coding process, as a new initial block and by submitting each quadrant 15A, 15B, 15C and 15D of the sub-block 14 to the same procedure (FIG. 2c).
  • Each of these new sub-blocks 15 thus contains for instance 2 ⁇ 2 pixels.
  • the small sub-blocks 15 are coded by utilizing the VQ process, so that in the third codebook 23, which is composed of a number of classified codebooks, there is looked up a suitable third codevector 24.
  • the respective vector indices i, j of the original block 13 e.g.
  • Each third codebook 23 contains a relatively small number of codevectors, and their number is generally more limited than the size of the codebooks 17, 20 used at the preceding coding level.
  • the image that is coded and further transmitted at the transmitting end, is at the receiving end decoded by a simple reversed decoding process, which is illustrated as a block diagram in FIG. 4.
  • the combined codebook 24 of the receiving end contains codebooks that are identical with the codebooks 17, 20, 22 of the transmitting end.
  • the codevectors 24 of the lowest level codebooks 23 are used for the actual compilation of the image.
  • the image is compiled block by block in the reconstruction unit 25 and is transmitted further in order to be reconstructed of blocks into an image in a previously known fashion.
  • Each codevector 23 is completely defined by its own vector index k in the particular codebook denoted by the classifying vector index j, i.
  • FIGS. 3 and 4 illustrate, in blocks marked with dotted lines, how coding and decoding is improved by means of averaging.
  • the mean value m of initial blocks of for instance 8 ⁇ 8 pixels is calculated in the averaging unit 26, and is further separated from each transmittable block in the differentiation unit 27.
  • the mean value means the average of the gray tones in the pixels contained by each block.
  • the mean value m is transmitted as a digital message to the transmission path and further to the receiving end.
  • the block difference i.e. the difference blocks, are coded and respectively decoded according to the method of the invention.
  • the block according to the mean value is reconstructed in the mean value block reconstruction unit 30, and it is summed in the summing unit 29 to the difference block in decoded form.
  • FIG. 5 An application of the compression coding method of the invention is illustrated in FIG. 5.
  • the image block is first reconstructed in the sub-coder 16, 19 on the basis of the information coded so far.
  • the difference between the image block and the reconstructed image block, i.e. the difference block, is formed in the differentiation unit 28, 29.
  • the difference block contains the coding errors of the previous coding stage.
  • the method of the invention is applied to the difference block in the fashion described above, instead of the image block.
  • like reference numbers are used of the separate units of the coder as earlier in connection with FIG. 3.
  • the decoding is in principle realized as a reversed operation with respect to the coding process.
  • Averaging can also be applied to the above described method of the invention, as is apparent from FIG. 5.
  • the averaging unit 26 and the differentiation unit 27 are adjusted to the coding system prior to the coding arrangement proper of the above described process.
  • Decoding is in principle realized in similar fashion as in the application of FIG. 4.
  • the codebook is designed by using a large number of statistically representative training images.
  • the generally known LBG algorithm can be used While designing the codebook.
  • the codebooks 17, 20, 23 are created by proceeding hierarchically, so that while designing each classified codebook, the employed training image material is that part of the original training image material which belongs to the class in question.
  • the first codebook 17 for the uppermost level is designed in the standard way by using the whole set of training blocks of the initial block size (e.g. 8 ⁇ 8 pixels).
  • each of the codebooks 20 is designed by using blocks of the original training images, which are first quantized with the first codebook and then grouped into different sets according to the resulting vector indices. In each set, the blocks are then divided into four quadrants (i.e. sub-blocks, such as 4 ⁇ 4 pixels) to form four separate subsets A', B', C', D'. All of the sub-blocks in the subset A' have the same classifying vector and quadrant indices. The obtained subsets are used as training material for the said codebooks 20.
  • the above described process for designing the codebooks can be continued in similar fashion, by subdividing the already created subsets of training images further into smaller subsets by classification with the codebooks already worked out and by a further division into four quadrants. Each new subset is used as a training set for a new codebook at the next lower level.
  • the block at the lowest level is called a basic block, and let its size be N ⁇ N pixels.
  • the hierarchical levels of blocks are labeled by the index h. This index h runs from 1 (lowest level) to L (top level).
  • the quadrant in question is denoted by the label " ⁇ h , which may take any of the four values A, B, C and D (cf. FIG. 2).
  • the coding proceeds in a treelike manner.
  • level h the whole chain of previous quantization indices and quadrant labels is: (1)
  • the block at level h is then vector quantized using the codebook of the class c h .
  • the resulting index i h together with the class index c h defines the composite vector index: (2)
  • the quadrant labels are not transmitted, because they are implied by the synchronism between the coding unit of the transmitting .end and the decoding unit of the receiving end.
  • the value of index i h of a block at the level h is shared by its sub-blocks and all their descendants.
  • bitrate of the whole process down to the lowest level, counted as bits per basic block (N ⁇ N pixels) is: (3) ##EQU1## Since all the values of the vector indices i h are not equally probable, the actual information transfer rate can further be reduced by entropy coding.
  • the entropy can be calculated assuming a codeword of length: (4)
  • index i h in class C h is associated to index i h in class C h , where P(i h
  • k h be the composite index of a block at the lowest level.
  • the total amount of information for this block is then: (5) ##EQU2## where k h now means k h truncated after level h (cf. equation 2).
  • the average total bit rate (entropy) is then: (6) ##EQU3##
  • P(k 1 ) of the lowest level composite vector indices k 1 may be estimated from their occurrence probabilities in the training set coded by the method.
  • the compression coding method of the invention can be understood as a special classification process inside the initial image blocks (e.g. 8 ⁇ 8 pixels), combined with the actual coding phase--i.e. vector quantization of sub-blocks on the lowest level (e.g. 4 ⁇ 4 or 2 ⁇ 2 sub-blocks).
  • the compression coding method of the invention uses previous VQ results in a natural way. There is no need to find any special criterium to separate the different classes.
  • TSVQ tree search VQ
  • successive coding results branches of the tree
  • codebook determines the codebook to be used in the next stage.
  • An essential difference in between the method of the invention and the said tree search method is, that in TSVQ the block size is fixed and remains the same through the whole process, whereas the method of the invention makes an extra branching towards smaller and smaller sub-blocks at each stage.
  • TSVQ mainly addresses the problem of coding time while still needing a very large codebook for good quality results, i.e. a high-quality image, with larger block sizes.
  • An advantage of the method of the present invention is its speed, which is achieved by coding based on hierarchy levels. Another advantage is that the numbers of the codevectors of the codebooks can be maintained on a reasonable level because of the hierarchically decreasing block sizes employed to define the details of the image. Viewed from a different angle, it can be said that the compression coding method of the invention is able, for a good outcome, also to utilize large scale correlations within the picture by starting with a larger block size.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)
  • Compression, Expansion, Code Conversion, And Decoders (AREA)
  • Compression Of Band Width Or Redundancy In Fax (AREA)
US08/341,534 1992-05-20 1993-05-11 Method for image compression coding in an image transmission system Expired - Lifetime US5692012A (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
FI922300A FI92272C (fi) 1992-05-20 1992-05-20 Kuvansiirtojärjestelmän tiivistyskoodausmenetelmä
FI922300 1992-05-20
PCT/FI1993/000203 WO1993023956A1 (en) 1992-05-20 1993-05-11 Method for image compression coding in an image transmission system

Publications (1)

Publication Number Publication Date
US5692012A true US5692012A (en) 1997-11-25

Family

ID=8535322

Family Applications (1)

Application Number Title Priority Date Filing Date
US08/341,534 Expired - Lifetime US5692012A (en) 1992-05-20 1993-05-11 Method for image compression coding in an image transmission system

Country Status (6)

Country Link
US (1) US5692012A (ja)
JP (1) JP3515572B2 (ja)
AU (1) AU4070793A (ja)
CA (1) CA2136215A1 (ja)
FI (1) FI92272C (ja)
WO (1) WO1993023956A1 (ja)

Cited By (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5861923A (en) * 1996-04-23 1999-01-19 Deawoo Electronics Co., Ltd. Video signal encoding method and apparatus based on adaptive quantization technique
US5936673A (en) * 1995-05-26 1999-08-10 Intel Corporation Temporal tile staggering for block based video compression
US5987181A (en) * 1995-10-12 1999-11-16 Sharp Kabushiki Kaisha Coding and decoding apparatus which transmits and receives tool information for constructing decoding scheme
US5987456A (en) * 1997-10-28 1999-11-16 University Of Masschusetts Image retrieval by syntactic characterization of appearance
US6304605B1 (en) * 1994-09-13 2001-10-16 Nokia Mobile Phones Ltd. Video compressing method wherein the direction and location of contours within image blocks are defined using a binary picture of the block
US6345126B1 (en) * 1998-01-29 2002-02-05 Xerox Corporation Method for transmitting data using an embedded bit stream produced in a hierarchical table-lookup vector quantizer
US6404923B1 (en) * 1996-03-29 2002-06-11 Microsoft Corporation Table-based low-level image classification and compression system
US20030007557A1 (en) * 1996-02-07 2003-01-09 Sharp Kabushiki Kaisha Motion picture coding and decoding apparatus
US20030219167A1 (en) * 2002-05-23 2003-11-27 Petri Pikkarainen Method and system for forming HCVQ vector library
US6731807B1 (en) * 1998-09-11 2004-05-04 Intel Corporation Method of compressing and/or decompressing a data set using significance mapping
US6873738B2 (en) * 1995-10-02 2005-03-29 Sony Corporation Hierarchical image processor for encoding or decoding, and memory on the same chip
US20080055331A1 (en) * 2006-08-31 2008-03-06 Ati Technologies Inc. Texture compression techniques
US20090148059A1 (en) * 2007-12-10 2009-06-11 Sharp Kabushiki Kaisha Image processing apparatus, image display apparatus, image forming apparatus, image processing method and storage medium
US20090252431A1 (en) * 2007-09-07 2009-10-08 Microsoft Corporation Image Resizing for Web-based Image Search
US20140052440A1 (en) * 2011-01-28 2014-02-20 Nokia Corporation Coding through combination of code vectors
US20140372449A1 (en) * 2009-03-17 2014-12-18 Ebay Inc. Image-based indexing in a network-based marketplace
US20150030238A1 (en) * 2013-07-29 2015-01-29 Adobe Systems Incorporated Visual pattern recognition in an image
US9495386B2 (en) 2008-03-05 2016-11-15 Ebay Inc. Identification of items depicted in images
US10147134B2 (en) 2011-10-27 2018-12-04 Ebay Inc. System and method for visualization of items in an environment using augmented reality
US10210659B2 (en) 2009-12-22 2019-02-19 Ebay Inc. Augmented reality system, method, and apparatus for displaying an item image in a contextual environment
US10878489B2 (en) 2010-10-13 2020-12-29 Ebay Inc. Augmented reality system and method for visualizing an item
US10936650B2 (en) 2008-03-05 2021-03-02 Ebay Inc. Method and apparatus for image recognition services
US11170003B2 (en) 2008-08-15 2021-11-09 Ebay Inc. Sharing item images based on a similarity score
US11651398B2 (en) 2012-06-29 2023-05-16 Ebay Inc. Contextual menus based on image recognition
US11869053B2 (en) 2012-03-22 2024-01-09 Ebay Inc. Time-decay analysis of a photo collection for automated item listing generation

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3013698B2 (ja) * 1994-04-20 2000-02-28 松下電器産業株式会社 ベクトル量子化符号化装置と復号化装置
DE19743662A1 (de) * 1997-10-02 1999-04-08 Bosch Gmbh Robert Verfahren und Vorrichtung zur Erzeugung eines bitratenskalierbaren Audio-Datenstroms
FI112424B (fi) * 2001-10-30 2003-11-28 Oplayo Oy Koodausmenetelmä ja -järjestely
CN100344163C (zh) * 2004-06-16 2007-10-17 华为技术有限公司 视频编解码处理方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0280314A2 (en) * 1987-02-25 1988-08-31 Fuji Photo Film Co., Ltd. Image signal compressing method by vector quantization
WO1990012460A1 (en) * 1989-04-06 1990-10-18 Utah State University Foundation Vector quantizer codebook processing circuit
EP0433268A2 (en) * 1985-02-28 1991-06-19 Mitsubishi Denki Kabushiki Kaisha Interframe adaptive vector quantization encoding apparatus and video encoding transmission apparatus
EP0450664A1 (en) * 1983-09-06 1991-10-09 Mitsubishi Denki Kabushiki Kaisha Vector quantizer for picture information
US5468069A (en) * 1993-08-03 1995-11-21 University Of So. California Single chip design for fast image compression
US5521988A (en) * 1994-04-05 1996-05-28 Gte Laboratories Incorporated Vector transform coder with multi-layered codebooks and dynamic bit allocation

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0450664A1 (en) * 1983-09-06 1991-10-09 Mitsubishi Denki Kabushiki Kaisha Vector quantizer for picture information
EP0457362A1 (en) * 1983-09-06 1991-11-21 Mitsubishi Denki Kabushiki Kaisha Vector quantizer
EP0433268A2 (en) * 1985-02-28 1991-06-19 Mitsubishi Denki Kabushiki Kaisha Interframe adaptive vector quantization encoding apparatus and video encoding transmission apparatus
EP0280314A2 (en) * 1987-02-25 1988-08-31 Fuji Photo Film Co., Ltd. Image signal compressing method by vector quantization
WO1990012460A1 (en) * 1989-04-06 1990-10-18 Utah State University Foundation Vector quantizer codebook processing circuit
US5468069A (en) * 1993-08-03 1995-11-21 University Of So. California Single chip design for fast image compression
US5521988A (en) * 1994-04-05 1996-05-28 Gte Laboratories Incorporated Vector transform coder with multi-layered codebooks and dynamic bit allocation

Cited By (57)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6304605B1 (en) * 1994-09-13 2001-10-16 Nokia Mobile Phones Ltd. Video compressing method wherein the direction and location of contours within image blocks are defined using a binary picture of the block
US5936673A (en) * 1995-05-26 1999-08-10 Intel Corporation Temporal tile staggering for block based video compression
US6473463B2 (en) 1995-05-26 2002-10-29 Intel Corporation Temporal tile staggering for block based video compression
US6873738B2 (en) * 1995-10-02 2005-03-29 Sony Corporation Hierarchical image processor for encoding or decoding, and memory on the same chip
US7194136B2 (en) 1995-10-12 2007-03-20 Sharp Kabushiki Kaisha Decoding apparatus using tool information for constructing a decoding algorithm
US6681050B1 (en) 1995-10-12 2004-01-20 Sharp Kabushiki Kaisha Decoding apparatus using tool information for constructing a decoding algorithm
US20040114818A1 (en) * 1995-10-12 2004-06-17 Sharp Kabushiki Kaisha Decoding apparatus using tool information for constructing a decoding algorithm
US20040101202A1 (en) * 1995-10-12 2004-05-27 Sharp Kabushiki Kaisa Decoding apparatus using tool information for constructing a decoding algorithm
US7167590B2 (en) 1995-10-12 2007-01-23 Sharp Kabushiki Kaisha Decoding apparatus using tool information for constructing a decoding algorithm
US6687409B1 (en) 1995-10-12 2004-02-03 Sharp Kabushiki Kaisha Decoding apparatus using tool information for constructing a decoding algorithm
US7340102B2 (en) 1995-10-12 2008-03-04 Sharp Kabushiki Kaisha Decoding apparatus using tool information for constructing a decoding algorithm
US20070081732A1 (en) * 1995-10-12 2007-04-12 Takeshi Makiyama Decoding apparatus using tool information for constructing a decoding algorithm
US6310981B1 (en) 1995-10-12 2001-10-30 Sharp Kabushiki Kaisha Decoding apparatus using tool information for constructing a decoding algorithm
US5987181A (en) * 1995-10-12 1999-11-16 Sharp Kabushiki Kaisha Coding and decoding apparatus which transmits and receives tool information for constructing decoding scheme
US6678326B2 (en) * 1996-02-07 2004-01-13 Sharp Kabushiki Kaisha Motion picture coding and decoding apparatus
US20030007557A1 (en) * 1996-02-07 2003-01-09 Sharp Kabushiki Kaisha Motion picture coding and decoding apparatus
US6404923B1 (en) * 1996-03-29 2002-06-11 Microsoft Corporation Table-based low-level image classification and compression system
US5861923A (en) * 1996-04-23 1999-01-19 Deawoo Electronics Co., Ltd. Video signal encoding method and apparatus based on adaptive quantization technique
US5987456A (en) * 1997-10-28 1999-11-16 University Of Masschusetts Image retrieval by syntactic characterization of appearance
US6671416B2 (en) 1998-01-29 2003-12-30 Xerox Corporation Method for transmitting data using an embedded bit stream produced in a hierarchical table-lookup vector quantizer
US6345126B1 (en) * 1998-01-29 2002-02-05 Xerox Corporation Method for transmitting data using an embedded bit stream produced in a hierarchical table-lookup vector quantizer
US6731807B1 (en) * 1998-09-11 2004-05-04 Intel Corporation Method of compressing and/or decompressing a data set using significance mapping
US7024048B2 (en) 1998-09-11 2006-04-04 Intel Corporation Method of compressing and/or decompressing a data set using significance mapping
US20040146208A1 (en) * 1998-09-11 2004-07-29 Pazmino Edward A Method of compressing and/or decompressing a data set using significance mapping
EP1367834A2 (en) * 2002-05-23 2003-12-03 Oplayo Oy Method and arrangement for forming HCVQ vector library
US20030219167A1 (en) * 2002-05-23 2003-11-27 Petri Pikkarainen Method and system for forming HCVQ vector library
EP1367834A3 (en) * 2002-05-23 2004-07-07 Oplayo Oy Method and arrangement for forming HCVQ vector library
US9418450B2 (en) * 2006-08-31 2016-08-16 Ati Technologies Ulc Texture compression techniques
US20080055331A1 (en) * 2006-08-31 2008-03-06 Ati Technologies Inc. Texture compression techniques
US11843793B2 (en) 2006-08-31 2023-12-12 Ati Technologies Ulc Texture decompression techniques
US11825106B2 (en) 2006-08-31 2023-11-21 Ati Technologies Ulc Texture decompression techniques
US11184628B2 (en) 2006-08-31 2021-11-23 Ati Technologies Ulc Texture decompression techniques
US10205956B2 (en) 2006-08-31 2019-02-12 Ati Technologies Ulc Texture compression techniques
US20090252431A1 (en) * 2007-09-07 2009-10-08 Microsoft Corporation Image Resizing for Web-based Image Search
US8081842B2 (en) * 2007-09-07 2011-12-20 Microsoft Corporation Image resizing for web-based image search
US8270772B2 (en) 2007-09-07 2012-09-18 Microsoft Corporation Image resizing for web-based image search
US20090148059A1 (en) * 2007-12-10 2009-06-11 Sharp Kabushiki Kaisha Image processing apparatus, image display apparatus, image forming apparatus, image processing method and storage medium
US8428395B2 (en) * 2007-12-10 2013-04-23 Sharp Kabushiki Kaisha Image processing apparatus, image display apparatus, image forming apparatus, image processing method and storage medium
US10936650B2 (en) 2008-03-05 2021-03-02 Ebay Inc. Method and apparatus for image recognition services
US11727054B2 (en) 2008-03-05 2023-08-15 Ebay Inc. Method and apparatus for image recognition services
US9495386B2 (en) 2008-03-05 2016-11-15 Ebay Inc. Identification of items depicted in images
US11694427B2 (en) 2008-03-05 2023-07-04 Ebay Inc. Identification of items depicted in images
US10956775B2 (en) 2008-03-05 2021-03-23 Ebay Inc. Identification of items depicted in images
US11170003B2 (en) 2008-08-15 2021-11-09 Ebay Inc. Sharing item images based on a similarity score
US9600497B2 (en) * 2009-03-17 2017-03-21 Paypal, Inc. Image-based indexing in a network-based marketplace
US20140372449A1 (en) * 2009-03-17 2014-12-18 Ebay Inc. Image-based indexing in a network-based marketplace
US10210659B2 (en) 2009-12-22 2019-02-19 Ebay Inc. Augmented reality system, method, and apparatus for displaying an item image in a contextual environment
US10878489B2 (en) 2010-10-13 2020-12-29 Ebay Inc. Augmented reality system and method for visualizing an item
US20140052440A1 (en) * 2011-01-28 2014-02-20 Nokia Corporation Coding through combination of code vectors
US10628877B2 (en) 2011-10-27 2020-04-21 Ebay Inc. System and method for visualization of items in an environment using augmented reality
US11113755B2 (en) 2011-10-27 2021-09-07 Ebay Inc. System and method for visualization of items in an environment using augmented reality
US10147134B2 (en) 2011-10-27 2018-12-04 Ebay Inc. System and method for visualization of items in an environment using augmented reality
US11475509B2 (en) 2011-10-27 2022-10-18 Ebay Inc. System and method for visualization of items in an environment using augmented reality
US11869053B2 (en) 2012-03-22 2024-01-09 Ebay Inc. Time-decay analysis of a photo collection for automated item listing generation
US11651398B2 (en) 2012-06-29 2023-05-16 Ebay Inc. Contextual menus based on image recognition
US9141885B2 (en) * 2013-07-29 2015-09-22 Adobe Systems Incorporated Visual pattern recognition in an image
US20150030238A1 (en) * 2013-07-29 2015-01-29 Adobe Systems Incorporated Visual pattern recognition in an image

Also Published As

Publication number Publication date
AU4070793A (en) 1993-12-13
FI92272B (fi) 1994-06-30
CA2136215A1 (en) 1993-11-25
FI92272C (fi) 1994-10-10
JP3515572B2 (ja) 2004-04-05
FI922300A0 (fi) 1992-05-20
WO1993023956A1 (en) 1993-11-25
FI922300A (fi) 1993-11-21
JPH07506707A (ja) 1995-07-20

Similar Documents

Publication Publication Date Title
US5692012A (en) Method for image compression coding in an image transmission system
US5764807A (en) Data compression using set partitioning in hierarchical trees
US5298896A (en) Method and system for high order conditional entropy coding
JP4264492B2 (ja) イメージをデコードする装置
US6157326A (en) Method of and device for coding a digital information signal
US5398069A (en) Adaptive multi-stage vector quantization
CN1155221C (zh) 编码方法及系统以及译码方法及系统
JPH07502632A (ja) ハフマンコードの復号化のための回路
US5313534A (en) Image compression method
JPH0695645B2 (ja) 符号化装置および符号化方法ならびに復号化装置および復号化方法
JP2000507754A (ja) 組込み形コーディングを有するテーブルベースの圧縮
US6614939B1 (en) Image compression apparatus and decoding apparatus suited to lossless image compression
JP3341962B2 (ja) 可変長復号器及び可変長符号値を復号化する方法
CN110392260B (zh) 一种低成本的联合图像压缩传输方法
EP1225543A2 (en) HVQ-based filtering method
Verma et al. A neural network based technique for data compression
US5369503A (en) Method of picture compression by auto-organisation of a neural network
Markas et al. Image compression methods with distortion controlled capabilities
Kamal Iteration free fractal image compression for color images using vector quantization, genetic algorithm and simulated annealing
US20090326935A1 (en) Method of treating voice information
EP1359755B1 (en) A coded information signal
Torres et al. Improvements on stochastic vector quantization of images
JP2693557B2 (ja) 符号化装置
Nahm et al. SAR image data compression algorithm for clipping service applications
US20020094126A1 (en) Fractal image compression

Legal Events

Date Code Title Description
AS Assignment

Owner name: VALTION TEKNILLINEN TUTKIMUSKESKUS, FINLAND

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:VIRTAMO, JORMA;VALLI, SEPPO;REEL/FRAME:008672/0850

Effective date: 19970731

STCF Information on status: patent grant

Free format text: PATENTED CASE

FPAY Fee payment

Year of fee payment: 4

FEPP Fee payment procedure

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

FPAY Fee payment

Year of fee payment: 8

AS Assignment

Owner name: PALUS A16, LLC, DELAWARE

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:VALTION TEKNILLINEN TUTKIMUSKESKUS;REEL/FRAME:020376/0477

Effective date: 20071025

FEPP Fee payment procedure

Free format text: PAYER NUMBER DE-ASSIGNED (ORIGINAL EVENT CODE: RMPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

FEPP Fee payment procedure

Free format text: PAYER NUMBER DE-ASSIGNED (ORIGINAL EVENT CODE: RMPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

FPAY Fee payment

Year of fee payment: 12